Currently we are storing saved queries for all namespaces in a single directory, where each file corresponds to a query and is named as uuid.json.
 
In order to load recent/favorite queries for a namespace, we have to read all of them and then filter by namespace. For users that have lots of queries, we have to read all of them which has led to the performance issues.
 
Even though we limit 30 queries per namespace, for users that deal with lots of dbs/collections, the number of (all) saved queries will continue to grow. As an improvement, it would be nice to folder them based on the namespace and then reads will be limited to the namespace folder - making it very fast.